Proc. Nati. Acad. Sci. USAVol. 84, pp. 6775-6779, October 1987Cell Biology

Purification of a membrane-derived human erythroid growth factor(burst-promoting activity/erythropoiesis/colony-forming units)

LAURIE FELDMAN*, CARL M. COHEN, MARY ALICE RIORDAN, AND NICHOLAS DAINIAKDepartments of Medicine and Biomedical Research, St. Elizabeth's Hospital and the Department of Medicine, Tufts University School of Medicine,Boston, MA 02135

Communicated by Eugene P. Cronkite, June 1, 1987

ABSTRACT We have purified erythroid burst-promotingactivity (BPA) from human lymphocyte plasma membranes bydetergent extraction followed by gel-filtration, ion-exchange,and hydroxylapatite chromatography. BPA is a heat-stableintegral membrane glycoprotein of Mr 28,000 by gel filtrationwhose activity is eluted from NaDodSO4/polyacrylamide gelsas a broad band at Mr 25,000-29,000. The growth stimulatorappears to be erythroid-specific, stimulating proliferation ofthe human erythroid burst-forming unit (BFU-E) by up to600% of control values when tested in serum-free bone marrowculture. In contrast, it is devoid of granulocyte/macrophagecolony-stimulating factor activity and has a negligible effect onthe formation of human megakaryocyte and mixed hemato-poietic colonies. Polyclonal anti-lymphocyte membrane IgG,which neutralizes BPA expression in culture, completely ab-sorbs BPA from all lymphocyte-derived sources [solubilizedlymphocyte plasma membranes, membrane-containing vesiclesshed into lymphocyte conditioned medium (LCM) and solublevesicle-free LCM supernatants], suggesting that soluble andmembrane-derived lymphocyte BPA are antigenically related.This membrane glycoprotein may be an important mediator ofproximal cellular interactions that are known to promoteerythropoiesis in vitro.

In vitro differentiation of hematopoietic, multipotential stemcells along the erythroid pathway requires that erythropoietinand/or erythroid burst-promoting activity (BPA) be added tothe culture medium (1-3). BPA is released spontaneously intoliquid culture medium from circulating and bone marrowlow-density mononuclear cells (4-7). We have previouslydemonstrated that BPA is expressed by lymphocyte plasmamembranes and by plasma membrane-derived vesicles thatare shed into lymphocyte conditioned medium (LCM) (7). Inorder to study the role of plasma membrane components ingrowth control as well as to investigate the relationship ofsoluble and membrane-derived growth factors in erythropoi-esis, we have purified BPA from the plasma membranes ofnormal, unstimulated peripheral blood lymphocytes. BPA isan integral membrane protein of Mr 28,000 on gel filtrationwhose activity comigrates with a single band on NaDod-S04/PAGE under reducing conditions. BPA stimulateserythroid burst formation in serum-free human bone marrowculture and exhibits no apparent synergistic interactions withother growth factors that we have tested. BPA is devoid ofgranulocyte/macrophage growth-factor activity (granulo-cyte/macrophage colony-stimulating factor, GM-CSF) andhas negligible stimulatory activity for megakaryocyte ormixed hematopoietic colony formation. Accordingly, thismembrane protein appears to be distinct from panspecific ormultilineage hematopoietic growth factors that have beendescribed previously (8, 9).

MATERIALS AND METHODSPreparation of Lymphocyte Plasma Membranes and Con-

ditioned Medium. Freshly drawn plateletpheresis residues,obtained from the Dana-Farber Cancer Institute, were frac-tionated by sedimentation in Ficoll-Paque (Pharmacia) asdescribed (7). Mononuclear cells were depleted of residualplatelets by the method of Perper et al. (10), and monocyteswere removed by adherence to plastic tissue culture flasks(Costar, Cambridge, MA) for 90 min (twice). Nonadherentcells (>97% lymphocytes by cytochemical and histochemicalidentification) were incubated at 5 x 106 per ml in serum-freealpha minimal essential medium (GIBCO) for production ofLCM. After overnight incubation at 370C in humidified 5%C02/air, cells were separated from the LCM by centrifuga-tion at 500 x g for 10 min. LCM was stored at 40C for furtheruse. The resultant lymphocytes (>98% viable by trypan blueexclusion) were washed twice in phosphate-buffered saline(PBS: 5 mM sodium phosphate/150 mM NaCl; pH 7.6 unlessotherwise indicated), and plasma membranes were isolatedby differential centrifugation and sucrose gradient fraction-ation according to a modification (11) of the method ofJett etal. (12). Purified plasma membranes were stored at -70'C inPBS. LCM was fractionated into supernatants and mem-brane-vesicle containing pellets by centrifugation at 40,000 Xg for 30 min. Pellets were resuspended in PBS and stored at-70'C, while supernatants were first subjected to a 30-50%(NH4)2SO4 cut and then stored at -70'C in PBS.Assay for Human Erythroid BPA. Approximately 0.5 ml of

spiculated bone marrow was aspirated from the posterior iliaccrests of normal volunteers. The cells were placed in Eagle'sminimal essential medium (GIBCO) containing heparin andwere separated by sedimentation in Ficoll-Paque. Mononu-clear cells layering at the interface were separated, washed inalpha medium and cultured at 6 x 105 per ml in serum-freefibrin clots as described (13). Cultures were supplementedwith Iscove's modified Dulbecco's medium (GIBCO), puri-fied human serum albumin, transferrin, ferric chloride, andeither sheep step-III erythropoietin (Connaught Laborato-ries, Willowdale, ON) or recombinant human erythropoietin(70,000-80,000 units/mg; Amgen Biologicals, ThousandOaks, CA) at 2.0 units/ml. Test plates contained 10%(vol/vol) LCM or its fractions, lymphocyte plasma mem-branes or fractions thereof, or NCTC-109 medium (controls).Cultures were maintained for 12 days at 370C in humidified5% C02/air and, following harvest, the fibrin clots werestained with benzidine and hematoxylin. Erythroid burstscontaining 250 benzidine-positive cells were scored under amicroscope. The amount of BPA present in experimental

Abbreviations: BPA, burst-promoting activity; CFU, colony-form-ing unit(s); CFU-E, -GM, -GEMM, and -Meg, erythroid, granulo-cyte/macrophage, multipotential, and megakaryocyte CFU, respec-tively; BFU-E, erythroid burst-forming unit(s); CSF, colony-stimu-lating factor; LCM, lymphocyte conditioned medium.*To whom reprint requests should be addressed at: Department ofBiomedical Research, St. Elizabeth's Hospital, 736 CambridgeStreet, Boston, MA 02135.

6775

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

6776 Cell Biology: Feldman et al.

plates was determined relative to that in control culturescontaining NCTC-109 ("100% growth").The mean ± SEM of quadruplicate determinations for each

test point was calculated, and data sets were compared by thetwo-sample ranks test of Wilcoxon and White (14).

Additional Culture Systems. The capacity of chromato-graphically purified BPA (see below) to support the prolif-eration of granulocyte/macrophage progenitors (CFU-GM),granulocyte/erythroid/macrophage/megakaryocyte progen-itors (CFU-GEMM), and megakaryocyte progenitors (CFU-Meg) was determined in human marrow cultures maintainedfor 12-14 days under serum-free conditions (13). Granulo-cyte/macrophage cultures were established with and without25 units of human colony-stimulating factor (CSF; Hyclone,Logan, UT) per ml. Megakaryocyte cultures were estab-lished with modification of the serum-free conditions accord-ing to Hoffman et al. (15) in the presence and absence ofpurified megakaryocyte CSF (16).

Extraction of Membrane-Derived BPA. Lymphocyte plas-ma membranes or LCM vesicles were suspended to 1 mg/mlin 5 mM Hepes, pH 7.4/2 mM dithiothreitol/0.2 mM phenyl-methylsulfonyl fluoride. Aliquots were mixed with equalvolumes of 0.2 M NaOH; 0.2 mM EDTA (pH 8.0); 2 M KCI;5 mM Hepes (pH 7.4); PBS; or 60 mM octyl ,3-D-gluco-pyranoside/5 mM Hepes, pH 7.4. Samples were incubated at0C for 30 min and then centrifuged at 100,000 x g for 30 minto separate solubilized proteins from residual insoluble mem-branes. Both supernatants and pellets (resuspended in PBS)were dialyzed extensively against PBS prior to bioassay.Chromatographic Fractionation of BPA. Procedures were

carried out at 4TC, unless indicated, and all buffers contained0.1% polyethylene glycol (PEG) 6000 (Sigma) to preventnonspecific sticking of proteins. Solubilized BPA-containingplasma-membrane or LCM-vesicle fractions or (NH4)2SO4-fractionated LCM supernatants were applied to 1.6 x 90-cmcolumns of Sephacryl S-300 (Pharmacia) and eluted with PBS(with or without 2 mM dithiothreitol). Individual fractionswere sterilized using 0.45-Ium filters prior to bioassay. Thosefractions demonstrating BPA in culture were pooled andconcentrated by lyophilization or by an Amicon Ultrafiltra-tion Stir Cell (Amicon) fitted with a YM10 membrane.Aliquots were dialyzed against 5 mM sodium phosphate (pH8.0) and applied to 5-ml columns of hydroxylapatite (Calbio-chem) or 1-ml columns of DE-52 DEAE-cellulose (What-man). For hydroxylapatite columns, elution was at roomtemperature with a step gradient from 5 mM to 100 mMsodium phosphate (pH 8.0). DE-52 columns were flushed atroom temperature with 5 mM sodium phosphate (pH 8.0),followed by stepwise elution with 0.01-1.0 M NaCl in 5 mMsodium phosphate (pH 8.0). Active fractions were pooled,concentrated, and dialyzed against PBS for storage. Alter-natively, following concentration, material was dialyzed into5 mM sodium phosphate (pH 8.0) and applied to the oppositecolumn (i.e., active material from DE-52 was applied tohydroxylapatite and vice versa). Finally, in some cases,active fractions were applied to a Bio-Sil TSK-250 HPLCcolumn (Bio-Rad) and eluted at room temperature with PBS(pH 7.0). After chromatographic fractionation, BPA wasstored at 4°C in PBS (pH 7.6).NaDodSO4/PAGE and Electroelution of BPA. Protein sam-

ples were electrophoresed in NaDodSO4 polyacrylamide gelsaccording to Laemmli (17), except that slab gels were used.Gels either were silver-stained by the method of Wray et al.(18) or, in the case of samples that were iodinated with'25I-labeled Bolton-Hunter reagent (19) (New England Nu-clear), were fixed, dried, and exposed for autoradiography.To localize unlabeled BPA in unstained gels, individual laneswere cut from the gel and sliced into 1-cm or 0.5-cm piecesthat were then homogenized by forcing them through a 10-ml

syringe. Homogenized samples were loaded into the samplewells (cathode) of an ISCO electrophoretic concentrator(ISCO) and electrophoretically eluted onto glycerol cushionsin the collection chambers (anode) at 150 V for 2 hr, with 10mM Tris acetate (pH 8.8) as the electrode buffer. Elutedproteins were dialyzed extensively against PBS containing0.1% PEG 6000 and were filter-sterilized for use in culture.Immunochemical Techniques. Antisera were prepared

against lymphocyte plasma membranes and against octylglucoside-solubilized membrane proteins (11). IgG fractionswere obtained by (NH4)2SO4 fractionation, followed byDEAE-cellulose chromatography. Pre- and postimmune IgGwere tested for crossreactivity with the immunizing antigensas well as with LCM fractions as described (11).For immunoadsorption, BPA-containing samples were

incubated overnight at 40C with increasing amounts of eitherpre- or postimmune antimembrane IgG. An amount of Pan-sorbin (Calbiochem) calculated to have an adsorption capac-ity 4-fold in excess of the maximum amount of IgG used wasadded and incubated for 15 min at room temperature beforebound IgG was removed by centrifugation. Adsorbed super-natants were then filter-sterilized and assayed for BPA inserum-free culture.

Protein Determination. Protein was estimated either by themethod of Lowry et al. (20) or by the fluorescamine method(21), using bovine serum albumin as a standard.

RESULTS

Purification of Lymphocyte Plasma Membrane-AssociatedBPA. Lymphocyte plasma membranes or plasma membrane-derived LCM vesicles were extracted with a variety ofreagents as described above. Of the reagents tested, only thedetergent octyl glucoside solubilized BPA from the mem-brane, suggesting that BPA is an integral membrane protein.Sequential extraction of membranes with 0.1 M NaOH,which extracts S50% of total protein while leaving BPAassociated with the insoluble membrane fraction, followed by30mM octyl glucoside resulted in >5-fold enrichment of BPAin the octyl glucoside supernatant.

Fractionation of octyl glucoside-solubilized proteins onSephacryl S-300 resulted in a chromatographic profile where-in the bulk of UV-absorbing protein was eluted in the voidvolume, and BPA, as measured by the activity of elutedfractions in serum-free marrow culture, was eluted with a Mrof 28,000 (Fig. 1). Although the inclusion of 2 mM dithio-threitol in the column buffer decreased slightly the size of thevoid-volume protein peak, it was without apparent effect onthe activity profile of the column.When BPA-containing fractions from the gel-filtration

column were applied to DE-52 in 5 mM sodium phosphate(pH 8.0) and eluted with a salt gradient, all growth-promotingactivity was eluted between 0.15 M and 0.25 M NaCI. Whenthis activity was concentrated, applied to a hydroxylapatitecolumn, and eluted with increasing phosphate concentra-tions, BPA was eluted at 20 mM phosphate (Fig. 2a).Reapplication of the active fractions to DE-52, under condi-tions identical to those used previously, resulted in elution ofBPA sharply at 0.2 M NaCI (Fig. 2b). Finally, when the 0.2M NaCl eluant from DE-52 chromatography was concentrat-ed and applied to a TSK-250 HPLC column in PBS (pH 7.0),BPA was eluted with a Mr of 28,000 (Fig. 2c). The chromato-graphic behavior of membrane-derived BPA was indistin-guishable from that of BPA derived from membrane vesicle-free LCM supernatant.NaDodSO4/PAGE. After electrophoresis of column-puri-

fled BPA, silver staining failed to reveal any stained bands.However, when proteins were prelabeled with 125I Bolton-Hunter reagent, subsequent autoradiography of the dried gelrevealed that the majority of radiolabel migrated in the

Proc. Natl. Acad. Sci. USA 84 (1987)

Proc. Natl. Acad. Sci. USA 84 (1987)

V ° i40

20 30 40 50 60 70 80Fraction

FIG. 1. Sephacryl S-300 fractionation of octyl glucoside-extract-ed lymphocyte plasma membranes. Solubilized membrane protein (2ml) was applied to a 1.6 x 90-cm column and eluted with PBS (pH7.6). Absorbance at 280 nm was measured (e), and individualfractions were assayed for growth-promoting activity on erythroidprogenitors (erythroid burst-forming units, BFU-E) in serum-freemarrow culture (0). Control growth (100%o) represents that in platescontaining NCTC-109 medium. Column calibration [void volume(V.) and elution positions of molecular weight standards (M, x 10-3)]is shown by arrows.

molecular weight range 25,000-29,000 (Fig. 3a). The broadnature of the BPA "band" even on a tightly crosslinked gel(25:1 acrylamide/N,N'-methylenebisacrylamide weight ra-tio) is consistent with the charge heterogeneity of glycopro-teins (22).

Electroelution of unlabeled BPA from 1-cm slices ofunstained gels indicated that all BPA was eluted at Mr25,000-29,000 (Fig. 3b), corresponding to the position of thelabeled material (Fig. 3a). For simplicity, only the molecularweight range(s) of interest are depicted in Fig. 3b; however,the entire gel was assayed for bioactivity. Those fractions notshown were totally devoid of BPA. Similar results were

FIG. 3. NaDodSO4/PAGE and electroelution of column-purifiedBPA. (a) Column-purified BPA was iodinated with 125I Bolton-Hunter reagent (19) and applied directly to an 11% polyacrylamidegel (17). The dried gel was exposed for autoradiography using KodakX-Omat paper. (b) After electrophoresis of unlabeled samples ofBPA from lymphocyte plasma membranes (PM) or from particulateor soluble fractions ofLCM [LCM(P) and LCM(S), respectively], theunfixed gel was sliced into 1-cm slices and proteins wereelectroeluted as described in Materials and Methods. Aliquots ofeluted protein were assayed in serum-free marrow culture.

obtained when unlabeled gels, composed of 10%, 12.5%, or15% acrylamide, were electroeluted (0.5- or 1-cm slices) andsubjected to bioassay.

Properties of Purified BPA. Column-purified BPA wasstable to heat treatment at 100'C for 10 min. Because thisresult was inconsistent with our earlier published results (7)suggesting that LCM subjected to 560C for 2 hr lost BPA, weinvestigated the heat stability of the growth-factor activitymore systematically. Samples of unfractionated LCM super-

b

10 20 40 60Phosphate, mM

80 100

c

0

0.01 0.2 0.4 0.6 0.8 1.0NaCl, M

20 30 40 50 60 70Fraction

FIG.2. Chromatographic fractionation of BPA. (a) Hydroxylapatite chromatography. Active fractions from Sephacryl S-300 (Fig. 1),concentrated and dialyzed against 5 mM sodium phosphate (pH 8.0), were applied to a 5-ml column of hydroxylapatite and eluted stepwise with5-100 mM sodium phosphate (pH 8.0). BPA was eluted with 20 mM phosphate. (b) Anion-exchange chromatography. Active fraction from a,concentrated and dialyzed against 5 mM sodium phosphate (pH 8.0), was applied to a 1-ml column of DE-52. Column was flushed with the samebuffer (o), and bound material was eluted stepwise with 0.01-1.0 M NaCl in 5 mM sodium phosphate (pH 8.0). BPA was eluted with 0.2 M NaCI.(c) HPLC. Active fraction from b was concentrated and dialyzed against PBS (pH 7.0). A 200-pi sample was applied to a Bio-Sil TSK-250 columnand eluted with the same buffer. BPA was eluted in a position corresponding to M, 28,000. Column calibration is shown by arrows.

0.6-

0.4-

0.2

ba

My' x 10 '66-

4-36-29.5

L-

0C)

.5

0C)

uo

24-

20()-

14-

34 29 25 22Mr X lo-,

a400r

300kto4-0

0C)

m200 k

100l

Cell Biology: Feldman et al. 6777

To00

6778 Cell Biology: Feldman et al.

natant or of partially purified BPA from either LCM super-natant or extracted lymphocyte plasma membranes (all withand without 2 mM dithiothreitol) were heated to either 560Cfor 2 hr or 100'C for 10 min, cooled on ice, and tested forBPA. The results in Table 1 indicate that regardless of thepresence or absence of dithiothreitol in the sample, theactivity of column-purified BPA was stable to treatment ateither 560C or 100TC. In contrast, in the absence of dithio-threitol, LCM lost virtually all activity when heated to eitherexperimental temperature for the times indicated. The find-ings are consistent with the possibility that heat inactivationofBPA in unfractionated LCM is mediated by contaminatingproteases or oxidants in the conditioned medium, and thatfractionation on Sephacryl S-300 is sufficient to separatethese contaminants from BPA.

Stability of BPA is further demonstrated by the ability torecover biologically active growth factor following NaDod-S04/PAGE and by the ability to store growth-factor prepa-rations for periods of up to a year at 40C or -70'C.Throughout chromatographic purification, membrane-de-

rived BPA can be adsorbed from solution by anti-lymphocytemembrane IgG. Therefore, although present on the lympho-cyte surface in minute quantities, membrane-associated BPAis highly immunogenic. The data in Fig. 4 indicate that BPAfrom octyl glucoside solubilization or from Sephacryl S-300or hydroxylapatite fractionation of membranes is completelyadsorbed to anti-membrane IgG. Additionally, SephacrylS-300-fractionated BPA from LCM supernatant is also ad-sorbed by anti-membrane IgG, indicating that soluble andmembrane-derived BPA are immunologically related. Incontrast, our previous work indicates that BFU-E-directedgrowth-promoting activities were not removed from mono-cyte conditioned medium (23), serum-free bovine aortaendothelial cell conditioned medium, serum-free Mo cellconditioned medium, or purified human CSF even at highantibody concentration, suggesting that these erythroidgrowth-regulatory factors are immunologically distinct fromlymphocyte-derived BPA.

Erythroid Specificity of BPA. Samples of column-purifiedor gel-eluted BPA were assayed in culture for their ability tostimulate the proliferation of granulocyte/monocyte (GM),megakaryocyte (Meg), or mixed (GEMM) progenitors. Incontrast to purified CSF, which stimulated CFU-GM to 866

10%o of control values, BPA failed to stimulate CFU-GMproliferation relative to cultures without added growth factor(Table 2). Multiple samples of BPA also failed to stimulateCFU-GEMM proliferation, and BPA did not stimulate CFU-Meg proliferation under conditions where 1 nM megakaryo-cyte CSF stimulated growth to 280 ± 5% of control values.

Table 1. Temperature stability of BPA

BFU-E, % control*

Sample DTT Untreated 56°C, 2 hr 100°C, 10 min

LCM(S) - 253 ± 10 110 ± 8 87 ± 2+ 247 ± 9 213 ± 8 247 ± 12

S-300LCM(S) - 263 ± 3 260 ± 6 274 ± 4

+ 256 ± 4 254 ± 2 246 ± 2S-300PM - 230 ± 7 220 ± 10 206 ± 5

+ 230± 5 200± 6 210± 7

Samples of LCM supernatant [LCM(S)], or of Sephacryl S-300-fractionated LCM(S) or extracted lymphocyte plasma membranes(PM), were treated in the presence or absence of 2 mM dithiothreitol(DTT). All samples were then cooled on ice and assayed in serum-free marrow culture.*Data are for quadruplicate determinations (±SEM). Control cul-tures (100% growth) contained 10% (vol/vol) NCTC-109 medium inplace of growth factor.

=2000

o100 _ -

25 50 100 200IgG, ug

FIG. 4. Immunoadsorption of BPA by anti-membrane IgG. Sam-ples (200 ,ul) of lymphocyte-derived BPA at various stages ofpurification were incubated with various amounts of anti-membraneIgG for 16 hr at 4°C prior to immunoprecipitation with excessPansorbin. Preadsorbed supernatants were then assayed in serum-free marrow culture for residual growth-factor activity. *, SephacrylS-300-fractionated LCM supernatant; o, octyl glucoside-solubilizedlymphocyte plasma membranes; o, Sephacryl S-300-fractionatedplasma membranes; A, plasma membranes fractionated over S-300,hydroxylapatite, and DE-52.

DISCUSSION

We have chromatographically purified human erythroid BPAfrom lymphocyte plasma membranes and have shown thatthis activity copurifies with a protein of Mr 28,000. BPA isexceptionally stable, resisting denaturation even after heattreatment at 100°C for 10 min. This behavior is consistentwith published data regarding heat stability of BPA (24) andother glycoprotein growth factors (25-27). Heat stability ofBPA has been used as a criterion for separating this activityfrom the colony-stimulating activity (CSA) with which itcopurifies from concanavalin A-stimulated T cells (24).Our purified BPA appears to be hematopoietic-lineage-

specific, failing to enhance the formation of granulocyte/macrophage, megakaryocyte, or mixed colonies. Further-more, it exhibits no apparent synergy with other growthfactors that we have tested. A more exhaustive survey usinghighly purified and/or recombinant growth factors will benecessary before we can conclude with confidence that BPAis indeed a unique growth factor. However, the data suggestthat our growth factor is a reasonable candidate for the

Table 2. Erythroid specificity of BPA% Control

LCM CFU- CFU- CFU-Purification step fraction GM* GEMM MegtS-300 Supernatant 115 ± 6 ND ND

Pellet 95 3 100 ± 8 89 ± 10.5Hydroxylapatite Supernatant 118 ± 6 115 ± 7.5 ND

Pellet 94 4 100 ± 1 120 ± 6PAGE Supernatant 97 ± 1 108 ± 6 ND

Pellet 107 ± 2 108 ± 3.5 82 ± 13

BPA from LCM supernatants or extracted vesicle-containingLCM pellets, purified through the steps indicated, was assayed inserum-free marrow culture for their ability to support the prolifera-tion of granulocyte/macrophage (GM), mixed (GEMM), or mega-karyocyte (Meg) precursors. Data represent the mean ± SEM forquadruplicate determinations in comparison to control cultures(100% growth) lacking added growth factor(s). ND, not done.*Value for CSF (from Hyclone; 25 units/ml) was 866 ± 10%6.tValue for megakaryocyte CSF (from R. Hoffmann; 1 nM) was 280± 5%.

Proc. Natl. Acad. Sci. USA 84 (1987)

Proc. Natl. Acad. Sci. USA 84 (1987) 6779

erythroid BPA that acts on (one of) the earliest committederythroid progenitor and is distinct from the factor(s) knownto act on multi- or pluripotential stem cells. The bone marrowtarget-cell population used in these studies is heterogeneousand undoubtedly contains a spectrum of erythroid progeni-tors ranging from early (BFU-E) to late (CFU-E). It is likelythat erythroid progenitors bear different growth-factor re-ceptors and, hence, have a differential response to BPA.Experiments performed with highly enriched bone marrowstem-cell populations should help to clarify this issue.

It was important to distinguish our growth factor fromerythroid potentiating activity (EPA), previously purified byWestbrook et al. (28). Both BPA and EPA have Mr of 28,000by gel filtration, are stable to heat treatment at 100'C, andstimulate proliferation of BFU-E progenitors. However, wehave observed that EPA is not absorbed from solution byantibodies (data not shown) that completely remove BPAfrom all lymphocyte-derived sources (see Fig. 4). Addition-ally, we have found that EPA is inactivated from crude orpartially purified Mo cell conditioned medium by the additionof 2 mM dithiothreitol to the growth factor at 250, 560, or100'C, suggesting that disulfide bonds are required for main-tenance of activity. This is in contrast to BPA, which appearsunaffected by the presence of 2 mM dithiothreitol. Mostimportant, whereas EPA is produced by a T-lymphoblast cellline (26), our preliminary evidence indicates that whennormal unstimulated peripheral blood lymphocytes are frac-tionated on nylon wool columns, BPA is found exclusivelyassociated with the non-T-cell (i.e., nylon wool-adherent)fraction (29).Although it is strongly suspected that cellular interactions

are critical for hematopoietic cell proliferation, little is knownabout precisely how such communication occurs. Our findingthat cells exfoliate surface membrane components that ex-press BPA (7) raises the possibility that proximity to, but notnecessarily contact with, target progenitor cells may besufficient to exchange growth-regulatory signals. Here, wereport that the active factor expressed by plasma membranesis biochemically and immunologically similar to that associ-ated with shed membrane-derived vesicles as well as to thatrecovered in vesicle-free LCM supernatants. This raisesintriguing questions not only regarding extracellular traffick-ing of shed surface components but also regarding the originof soluble BPA. The role of the plasma membranes eitherdirectly or as a potential generator of soluble messengers incell-cell communication and growth control remains an areaof intense interest.

We wish to acknowledge Mr. Joe Connelly for performing HPLCruns, Ms. Joan Joos for the artwork, and Ms. Donna A. MacDonaldfor typing the manuscript. We thank Dr. Ronald Hoffman (Universityof Indiana) for performing assays on human megakaryocyte progen-itor cells and Dr. David Golde (UCLA) for providing us withserum-free Mo cell conditioned medium. This work was supported in

part by National Institutes of Health Grants AM31060 and AM27071to N.D.

1. Iscove, N. N. (1978) in Hemopoietic Cell Differentiation, eds.Golde, D. W., Cline, J. J., Metcalf, D. & Fox, C. F. (Aca-demic, New York), pp. 37-52.

2. Axelrad, A. A., McLeod, D. L., Suzuki, S. & Shreeve, M. M.(1978) in Differentiation ofNormal and Neoplastic Cells, eds.Clarkson, B., Marks, P. A. & Till, J. E. (Cold Spring HarborLaboratory, Cold Spring Harbor, NY), pp. 155-163.

3. Goodman, J. W., Hall, E. A., Miller, K. L. & Shinpock, S. G.(1985) Proc. Natl. Acad. Sci. USA 82, 3291-3295.

4. Aye, M. T. (1977) J. Cell Physiol. 91, 69-77.5. Porter, P. N., Ogawa, M. & Leary, A. G. (1980) Exp. Hema-

tol. 8, 83-88.6. Porter, P. N. & Ogawa, M. (1982) Blood 59, 1207-1212.7. Dainiak, N. & Cohen, C. M. (1982) Blood 60, 583-594.8. Schrader, J. W. (1986) Annu. Rev. Immunol. 4, 205-230.9. Donahue, R. E., Emerson, S. G., Wang, E. A., Wong, G. G.,

Clark, S. C. & Nathan, D. G. (1985) Blood 66, 1479-1481.10. Perper, R. J., Zee, T. W. & Mickelson, M. M. (1978) J. Lab.

Clin. Med. 72, 842-848.11. Dainiak, N., Feldman, L. & Cohen, C. M. (1985) Blood 65,

877-885.12. Jett, M., Steed, T. M. & Jamieson, G. A. (1977) J. Biol. Chem.

252, 2134-2142.13. Dainiak, N., Kreczko, S., Cohen, A. M., Pannell, R. &

Lawler, J. (1985) Exp. Hematol. 13, 1073-1079.14. Goldstein, A. (1964) Biostatistics: An Introductory Text (Mac-

millan, New York), p. 62.15. Hoffman, R., Bruno, E., Yang, H., Sledge, G., Brandt, J. &

Beyer, G. (1986) Clin. Res. 34, 658A.16. Hoffman, R., Yang, H. H., Bruno, E. & Stravena, J. E. (1985)

J. Clin. Invest. 75, 1174-1182.17. Laemmli, U. K. (1970) Nature (London) 227, 680-685.18. Wray, W., Boulikas, T., Wray, V. P. & Hancock, R. (1981)

Anal. Biochem. 118, 197-203.19. Bolton, A. E. & Hunter, W. M. (1973) Biochem. J. 133,

529-539.20. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall,

R. J. (1951) J. Biol. Chem. 193, 265-275.21. Udenfriend, S., Stein, S., Bohlen, P., Dairmain, W., Leim-

gruber, W. & Weigele, M. (1972) Science 178, 871-872.22. Wong, G. G., Witek, J. S., Temple, P. A., Wilkens, K. M.,

Leary, A. C., Luxenberg, D. P., Jones, S. S., Brown, E. L.,Kay, R. M., Orr, E. C., Shoemaker, C., Golde, D. W., Kauf-man, R. J., Hewick, R. M., Wang, E. A. & Clark, S. C. (1985)Science 228, 810-815.

23. Feldman, L., Cohen, C. M. & Dainiak, N. (1986) Blood 67,1454-1459.

24. Sullivan, R., Hesketh, P. J., McCarroll, L. A. & Zuckerman,K. S. (1986) Exp. Hematol. 14, 659-667.

25. Ross, R. & Vogel, A. (1978) Cell 14, 203-210.26. Golde, D. W., Bersch, N., Quan, S. G. & Lusis, A. J. (1980)

Proc. Natl. Acad. Sci. USA 77, 593-5%.27. Gauwersky, C. E., Lusis, A. J. & Golde, D. W. (1982) Blood

59, 300-305.28. Westbrook, C. A., Gasson, J. C., Gerber, S. E., Selsted,

M. E. & Golde, D. W. (1984) J. Biol. Chem. 259, 9992-9995.29. Dainiak, N. & Cohen, C. M. (1985) Ann. N. Y. Acad. Sci. 459,

129-142.

Cell Biology: Feldman et al.